Cannabis dimensionality: dependence, abuse and consumption.

AIMS Genetic research on substance use disorders usually defines phenotypes as a binary diagnosis, resulting in a loss of information if the disorder is inherently dimensional. The DSM-IV criteria for drug dependence were based on a theoretically dimensional (linear) model. Considerable investigation has been conducted on DSM-IV alcohol criteria, but less is known about the dimensionality of DSM-IV cannabis criteria for abuse and dependence. The aim of this study is to assess whether DSM-IV cannabis dependence (including withdrawal) and abuse criteria fit a linear measure of severity and whether a consumption criterion adds linearly to severity. DESIGN/SETTING/PARTICIPANTS/MEASUREMENTS Participants were 8172 in the National Epidemiologic Survey on Alcohol and Related Conditions who had ever used cannabis. Wald statistics were used to test whether categorical, dimensional or hybrid forms best fit the data. We examined the following as criterion sets: (1) dependence; (2) dependence and abuse; and (3) dependence, abuse and frequency of use. Validating variables included family history of drug problems, early onset of cannabis use, and antisocial personality disorder. FINDINGS For cannabis dependence, no evidence was found for categorical or hybrid models; Wald tests indicated that models representing the seven DSM-IV dependence criteria as a linear severity measure best described the association between the criteria and validating variables. However, significant differences from linearity occurred after adding the four cannabis abuse criteria (p=0.03) and the use indicator (p=0.01) for family history and antisocial personality disorder. CONCLUSION With ample power to detect non-linearity, cannabis dependence was shown to form an underlying continuum of severity. However, adding abuse criteria, with and without a measure of consumption, resulted in a model that differed significantly from linearity for two of the three validating variables.